The present invention relates generally to data analysis and, more particularly, to reporting and comparison of data analysis results.
The method of reporting numerical laboratory test data, such as biological and other laboratory tests, has essentially remained unchanged since its modern inception, beginning in the first half of the twentieth century. The traditional method includes reporting a measured value (i.e., a test result) and its relevant set of normal values, known as a reference range. It is often inadequate to report only a measured value because different tests may have different respective reference ranges. Generally, all reference ranges include a set of two values with one value designated as an upper reference range limit and another designated as a lower reference range limit.
In the last quarter of the twentieth century the number of available laboratory tests has risen prodigiously and there are now many hundreds of numerically reported tests, each continuing to have its own unique set of reference ranges. This marked proliferation of data has offered an interpreter of the data an abundant variety of tests from which to conduct physiological as well as disease investigations. However, the sheer volume of available tests has also contributed to information overload. An interpreter typically attempts to remember hundreds of reference ranges when evaluating test data. For example, a single composite tabular listing of lab results on one biological entity (or non-biological entity) can include forty or more tests, all of which may have different reference ranges.
Another aspect of the interpretation and application of measured biological (or non-biological) laboratory data is the observation that a test value that falls within the reference range has variable significance depending on whether the measured value is near the upper limit, the lower limit, or the mean value of the reference range. The relative significance of a test has to be qualitatively assessed and committed to memory because it is not typically quantified on the traditional report. If multiple tests are simultaneously reported, an interpreter of the test data typically tries to retain in his/her memory the relative position of each measured value and make qualitative interpretive decisions among the tests utilizing mentally calculated relative positions in the reported test data. For example, one test may have a measured value two points below the upper reference range value and another test may have a measured value eight points below the upper reference range value. The interpreter may wish to know if one of these tests is at more risk for being abnormally elevated than the other test. A qualitative evaluation may be required because the number of points in the reference range for each of these tests may be different. The relative closeness of one value to the upper reference range (or the lower reference range for that matter) may be dependent on the number of units in the reference range. Table 1 below illustrates this situation.
In Table 1, the second column (MV=xe2x88x922) indicates a measured value two numbers less than the upper limit of the reference range. The third column (MV=xe2x88x928) indicates a measured value eight numbers less than the upper limit of the reference range. Each of the six measured values (MVs) in Table 1 are considered normal values because each lies within the reference range for a respective test. When measured values are viewed in the format of Table 1, which resembles traditional reporting formats, it may be difficult to determine which measured value is relatively greater than, or less than, any other measured value.
Consequently, if measured values that fall within a reference range are to be compared among the many different tests, then an interpreter should perform a qualitative analysis on each test and retain this information in memory for each test. If this type of mental calculation is not performed, then refinement in the application of measured values may not be possible and diagnostic information may be lost.
The concept of relative normalcy of a measured value that falls within a reference range is also applicable to measured abnormal values that are above or below a reference range. The same qualitative mental assessment is involved in determining the relative abnormalcy of an abnormal value. An example of this would be to determine whether a liver function test that is elevated ten points above the upper reference range is as qualitatively elevated as another liver function test that is also ten points above the upper reference range. Since these two tests may indicate different parts of the liver, it is reasonable to ask whether one part of the liver is more diseased than the other. This process of test comparison may become even more complex when the interpreter is attempting to assess a panel of many tests that relate to different organs or different diseases. This type of analysis is generally referred to as multiparametric analysis.
There have been attempts to present multiparametric test data from biological entities in a non-traditional format in order to enhance an interpreter""s perception of inter-test relationships and abnormal values. For example, U.S. Pat. No. 4,527,240 to Kvitash describes a process whereby measured patient values are transformed to units referred to as xe2x80x9cBalascopicxe2x80x9d units. Unfortunately, in data analysis according to Kvitash, the Balascopic units are plotted on an axial graph. These axial graphs may be somewhat difficult to use. Furthermore, the Balascopic process of Kvitash does not distinguish between test data reported as whole integers and decimals. Consequently, interpretative decisions that are made based on decimal values may be difficult to make with the Kvitash process. Another drawback of the Kvitash process is that it does not provide analytical variation associated with each measured value.
U.S. Pat. No. 5,541,854 to Yundt describes displaying conventional multi-level hematology quality control data (three levels) in a complex graphic form. Yundt is concerned with the presentation of tri-level quality control data and not with the presentation of measured unknown samples.
Statistical methods utilizing xe2x80x9cZ scoresxe2x80x9d to specify the relative frequency or probability of a random number in a normally distributed set of measurements are known. Unfortunately, Z scores are somewhat difficult to use to identify the relative value of one test result to another. Furthermore Z score techniques are somewhat limited because data beyond the maximum and minimum limits of normal distribution cannot be used.
In view of the above discussion, it is an object of the present invention to reduce much of the complexity associated with interpreting laboratory test data.
It is another object of the present invention to facilitate determining relative relationships of measured values from laboratory tests to respective reference ranges.
These and other objects are provided by systems, methods, and computer program products for determining relative normalcy or abnormalcy (or any other type of variation) of a plurality of test results, by transforming test results into respective unitized values and then graphically displaying each of the unitized values with a unitized reference range. Additionally, a unitized analytical variation of each of the respective unitized values may be determined and displayed.
Initially, a reference range for a test is unitized to a single number. A total number of possible test results within respective portions (e.g., halves) of the reference range is then determined. This is accomplished by determining a total number of possible test results within the reference range to produce a reference range spread, and then dividing the reference range spread in portions.
The fractional value of the plurality of test results in the respective portions of the reference range is then determined. The fractional value of the plurality of test results comprises a reciprocal of the total number of possible test results in the portion of the reference range. Each of the plurality of separately determined test results are then transformed into respective equilibrated values. This is accomplished by determining the equilibration point of the reference range and then determining a difference between the equilibration point and each of the plurality of test results. Each equilibrated value represents a relative position of a respective test result with respect to an equilibration point of the reference range.
Each of the equilibrated values is then transformed into respective unitized values by multiplying each respective equilibrated value with a respective fractional value of the plurality of test results in the portion of the reference range. Each unitized value represents relative normalcy or abnormalcy of a respective test result with respect to the unitized reference range.
According to the present invention, unitized values with the same numerical value indicate the same quantitative variation from any reference point(s) within the unitized reference range. According to the present invention, a unitized value of 1.5 for a glucose level within a patient and a unitized value of 1.5 for a sodium level within a patient will mean the same quantitative increase for each test. Furthermore, the present invention may allow an interpreter to recognize problems and undertake corrective actions sooner. By utilizing unitized values according to the present invention, an interpreter could more easily recognize that a unitized sodium value of 1.5 is more severe than a unitized glucose value of 1.1 and, therefore, take action to rectify the sodium level.
According to another embodiment of the present invention, unitized data can be graphically presented in a predefined order to produce patterns. These patterns can be compared, either manually or electronically, with other patterns representative of known medical conditions to quickly and easily assess the medical status of multiple body organs and/or systems of a patient.
The present invention may be applied to the interpretation of any type of laboratory test data, both biological and non-biological, and particularly where the test data is interpreted by referring to a reference range. Furthermore, the present invention may be applied to the interpretation of data measured from any entity, system, or process which is capable of being measured, including biological and non-biological entities. The present invention is particularly useful where multiparametric data is obtained from testing.